REPORT ON THE IMPACT OF SEATTLE S MINIMUM WAGE ORDINANCE ON WAGES, WORKERS, JOBS, AND ESTABLISHMENTS THROUGH 2015

Transcription

1 REPORT ON THE IMPACT OF SEATTLE S MINIMUM WAGE ORDINANCE ON WAGES, WORKERS, JOBS, AND ESTABLISHMENTS THROUGH 2015 The Seattle Minimum Wage Study Team 1 University of Washington July 2016 Daniel J. Evans School of Public Policy & Governance University of Washington Box Seattle, WA School of Social Work University of Washington Box Seattle, WA School of Public Health University of Washington Box Seattle, WA Suggested Citation: The Seattle Minimum Wage Study Team Report on the Impact of Seattle s Minimum Wage Ordinance on Wages, Workers, Jobs, and Establishments Through Seattle. University of Washington. 1 The team investigators are Jacob Vigdor, Mark C. Long, Jennifer Romich, Scott W. Allard, Heather D. Hill, Jennifer Otten, Robert Plotnick, Scott Bailey, and Anneliese Vance-Sherman. This report was principally authored by Long, Plotnick, Ekaterina Roshchina, Vigdor, and Hilary Wething, and had major intellectual and editing contributions from other members of the Seattle Minimum Wage Study Team. Any opinions expressed in this report are those of individual SMWS contributing investigators and not those of the University of Washington, the Washington Employment Security Department, or any supporting or contracted entity. To contact the Seattle Minimum Wage Study Team, mwage@uw.edu.

2 Executive Summary This report presents the short-run effects of the Seattle Minimum Wage Ordinance on the Seattle labor market. The Seattle Minimum Wage study team at the University of Washington analyzed administrative records on employment, hours, and earnings from the Washington Employment Security Department to address two fundamental questions: 1) How has Seattle s labor market performed since the City passed the Minimum Wage Ordinance, and particularly since the first wage increase phased in on April 1, 2015? 2) What are the short-run effects of the Minimum Wage Ordinance on Seattle s labor market? While quite similar at first glance, these two questions address very different issues and require very different methods to answer. The first question can be studied with a simple before/after comparison. Although the comparison is simple, it risks conflating the impact of the minimum wage with other local trends. Many things have happened in Seattle s labor market since June 2014, most of them having little or nothing to do with the minimum wage itself. The City has enjoyed steady expansion in tech sector employment, and a construction boom fueled by rising residential and commercial property prices. Even the weather a key determinant of economic activity in the Puget Sound region was favorable in 2015, with record-low precipitation in the early months of the $11 minimum wage. The before-after comparison can tell us the net impact of all these simultaneous trends, but this comparison cannot distinguish among them. Our second question the more important one for purposes of evaluating the policy aims to isolate the impact of the minimum wage from all the other regional trends seen over the same time period. Whereas the first question asks are we better off than we were when Seattle raised the minimum wage and requires only a simple comparison of yesterday to today, the second asks are we better off than we would have been if Seattle had not adopted a higher minimum wage? To answer it requires imagining how the local economy would look in absence of a Minimum Wage Ordinance. While it is impossible to directly observe what would have happened if no wage ordinance had been implemented, this report uses widely accepted statistical techniques to compare Seattle in its current state with the presence of the Minimum Wage Ordinance to an image of what Seattle might have looked like today if not for the Minimum Wage Ordinance. We take advantage of data going back to 2005 to build a model of the way Seattle s labor market typically works. We also take advantage of data on nearby regions that did not increase the minimum wage to better understand how other factors might have influenced what we observe in the City itself. 2

3 In this report, we present findings on wages, workers, jobs, and establishments. Our findings can be summarized as follows: Wages: The distribution of wages shifted as expected. The share of workers earning less than $11 per hour declined sharply. This decline began shortly after the ordinance was passed. However, similar declines were seen outside of Seattle, suggesting an improving economy may be the cause of the change in the distribution of wages. Low-Wage Workers: In the 18 months after the Seattle Minimum Wage Ordinance passed, the City of Seattle s lowest-paid workers experienced a significant increase in wages. The typical worker earning under $11/hour in Seattle when the City Council voted to raise the minimum wage in June 2014 ( low-wage workers ) earned $11.14 per hour by the end of 2015, an increase from $9.96/hour at the time of passage. The minimum wage contributed to this effect, but the strong economy did as well. We estimate that the minimum wage itself is responsible for a $0.73/hour average increase for low-wage workers. In a region where all low-wage workers, including those in Seattle, have enjoyed access to more jobs and more hours, Seattle s low-wage workers show some preliminary signs of lagging behind similar workers in comparison regions. The minimum wage appears to have slightly reduced the employment rate of low-wage workers by about one percentage point. It appears that the Minimum Wage Ordinance modestly held back Seattle s employment of low-wage workers relative to the level we could have expected. Hours worked among low-wage Seattle workers have lagged behind regional trends, by roughly four hours per quarter (nineteen minutes per week), on average. Low-wage individuals working in Seattle when the ordinance passed transitioned to jobs outside Seattle at an elevated rate compared to historical patterns. Seattle s low-wage workers did see larger-than-usual paychecks (i.e., quarterly earnings) in late 2015, but most if not all of that increase was due to a strong local economy. Increased wages were offset by modest reductions in employment and hours, thereby limiting the extent to which higher wages directly translated into higher average earnings. At most, 25% of the observed earnings gains around a few dollars a week, on average can be attributed to the minimum wage. Seattle s low-wage workers who kept working were modestly better off as a result of the Minimum Wage Ordinance, having $13 more per week in earnings and working 15 minutes less per week. 3

4 Jobs: Overall, the Seattle labor market was exceptionally strong over the 18 months from mid to the end of Seattle s job growth rate tripled the national average between mid-2014 and late This job growth rate outpaced Seattle s own robust performance in recent years. Surrounding portions of King County also had a very good year; the boom appears to fade with geographic distance. Job growth is clearly driven by increased opportunities for higher-wage workers, but businesses relying on low-wage labor showed better-than-average growth as well. For businesses that rely heavily on low-wage labor, our estimates of the impact of the Ordinance on the number of persistent jobs are small and sensitive to modeling choices. Our estimates of the impact of the Ordinance on hours per employee more consistently indicate a reduction of roughly one hour per week. Fewer hours per employee could reflect higher turnover rather than cutbacks in staffing. Reductions in hours are consistent with the experiences of low-wage workers. Establishments: We do not find compelling evidence that the minimum wage has caused significant increases in business failure rates. Moreover, if there has been any increase in business closings caused by the Minimum Wage Ordinance, it has been more than offset by an increase in business openings. In sum, Seattle s experience shows that the City s low-wage workers did relatively well after the minimum wage increased, but largely because of the strong regional economy. Seattle s low wage workers would have experienced almost equally positive trends if the minimum wage had not increased. Although the minimum wage clearly increased wages for this group, offsetting effects on low-wage worker hours and employment muted the impact on labor earnings. We strongly caution that these results show only the short-run impact of Seattle s increase to a wage of $11/hour, and that they do not reflect the full range of experiences for tens of thousands of individual workers in the City economy. These are average effects which could mask critical distinctions between workers in different categories. Our future work will extend analysis to 2016, when Seattle s minimum wage increased a second time and began to distinguish between businesses of different sizes and industries. It will also incorporate more detailed information about workers by linking employment records to other state databases. This will give us a greater capacity to answer key questions, such as whether the workers benefiting most from higher minimum wages are more likely to be living in poverty. We are also in the process of collecting additional survey information from Seattle businesses and conducting interviews with a worker sample tracked since early The next report, expected in September, will focus specifically on how the minimum wage has affected nonprofit organizations. 4

5 Table of Contents 1. Introduction Methodology, Data, and Outcomes Studied Impact of Seattle s Minimum Wage Ordinance on Wages Impact of Seattle s Minimum Wage Ordinance on Low-Wage Workers Impact of Seattle s Minimum Wage Ordinance on Jobs Impact of Seattle s Minimum Wage Ordinance on Establishments Conclusion and Discussion of Next Steps Appendix A. The Employment Security Department Data Appendix B: Detailed Discussion of Methodology Appendix D: Tables for Impact on Low-Wage Workers Appendix E: Tables for Impact on Jobs Appendix F: Tables for Impact on Establishments Appendix G: UW Minimum Wage Study Team Acknowledgments Thank you to the state of Washington s Employment Security Department for providing the Study Team access to data on which this report is based. Thanks to the Laura and John Arnold Foundation, the Russell Sage Foundation, and the City of Seattle for funding and support the Seattle Minimum Wage Study. Partial support for this study came from a Eunice Kennedy Shriver National Institute of Child Health and Human Development research infrastructure grant, R24 HD042828, to the Center for Studies in Demography & Ecology at the University of Washington. 5

6 1. Introduction This report uses administrative records from the state of Washington s Employment Security Department (ESD) to estimate the short-run impact of Seattle s Minimum Wage Ordinance on wages, workers, jobs, and establishments through the end of 2015, which was nine months after the ordinance went into effect on April 1, The ESD data, described in detail below, are gathered quarterly. While reading this report, it is useful to keep the following key dates (and their corresponding quarters) in mind: June, 2014 (End of 2 nd Quarter): Law was passed. April, 2015 (Beginning of 2 nd Quarter): Law went into effect, requiring the following wages be paid. $11.00 per hour for Schedule 1 employers (more than 500 employees in the U.S.) and Schedule 2 employers using guaranteed minimum compensation (e.g. tips and medical care) to reach $ $10.00 per hour for Schedule 2 employers not using guaranteed minimum compensation. 4th Quarter 2015: End of ESD records currently available. January, 2016: Minimum wage increased to as much as $13 (for Schedule 1 employers without medical benefits). It also is essential to note that this report estimates only the initial effects of the ordinance, examining the first nine months of a planned seven-year phase-in period. In Figure 1, we highlight, in orange, the period this report covers. Economic theory predicts the long-run adjustments to a regulatory policy change are likely to be greater than short-run adjustments. Prior research shows that in the long-run, certain industries affected by the minimum wage, such as the fast food industry, have more opportunity to relocate, change the composition of their workforce, or invest in technologies that reduce their need for labor. 2 Similarly, it is likely that any effects of higher earnings on health or family well-being would only emerge over the longer run. Additionally, the impacts that we find for this period, when the highest local minimum wage of $11 per hour only modestly exceeded the State s minimum wage of $9.47, may be quite different than the effects we will find following the City s increase to $13 per hour and above. This report will address the following questions: 1. What happened to the distribution of hourly wages? Does the change in the distribution match the expected effects of the Minimum Wage Ordinance? 2. What happened to low-wage workers wages, likelihood of remaining employed, hours worked, total earnings, and likelihood of remaining employed in Seattle? 2 See for example Daniel Aaronson, Eric French, and Isaac Sorkin, "Industry Dynamics and the Minimum Wage: A Putty-Clay Approach", Working Paper, 2016, 6

7 3. What happened to jobs, hours worked, and hours per job, particularly at establishments where a large share of employees worked for low wages? 4. What happened to the closure rate and opening rate of establishments in Seattle? For each of these questions, we will provide answers for what happened and for what is the estimated impact of the Minimum Wage Ordinance. The next section explains the data and methodology used to answer these questions. The four sections that follow discuss the impacts on wages, low-wage workers, jobs, and establishment closures and openings, respectively. Figure 1 Figure 1: Period Evaluated in This Report 3 Period evaluated in this report 3 Original source for figure, modified by the Study Team to include the highlighted oval: 7

8 2. Methodology, Data, and Outcomes Studied Methodology In addition to evaluating what happened in Seattle, we follow the conventions of modern policy evaluation research by asking the question what happened in Seattle, relative to what would have happened in the absence of an increased minimum wage? This perspective means that we will not simply look at basic labor market statistics before and after April 1, 2015; this might reveal what happened in Seattle but does not indicate what would have happened in the absence of an increased minimum wage. To take that additional step, we will attempt to construct an estimate of the counterfactual and to evaluate the difference between the observed and counterfactual outcomes. City stakeholders with an interest in understanding and speaking about the results of our analysis will be well served to review the distinction between what happened and what happened relative to what would have happened. Figure 2 graphically illustrates our approach, which is to decompose the observable change in outcomes into the portions explained by business as usual (Seattle relative to its own prior history), economic climate (given by the comparison region), and the remainder, which is our estimate of the causal impact of the minimum wage law. Figure 2 Figure 2: The Difference in Differences Strategy The most straightforward approach to evaluating the effects of the Minimum Wage Ordinance is analogous to simple before and after snapshots of labor market trends (i.e., the observable change ). That is, we examine what occurred in the period when the minimum wage was implemented and compare it to previous periods in Seattle s recent history. For example, we plot outcomes, notably the share of worker earning under $11, over time and look for changes in the level or trend of the outcomes following the introduction of the Minimum Wage Ordinance. This method, known as interrupted time series relies on the strong assumption that pre-ordinance time-trends would have continued if the ordinance had not passed. But there may be other contemporaneous changes that also affect the outcomes. For example, a sizeable change in national economic conditions could affect business and labor market outcomes in Seattle. We improve our estimates of the ordinance s impacts by benchmarking the Seattle experience to both Seattle s own prior history and to that of comparison regions. This approach, which is called 8

9 difference-in-differences, compares the difference in outcomes that Seattle experiences over time to the difference in outcomes of a valid comparison group. The change in outcomes in comparison regions serves as the counterfactual for the changes in Seattle. The method captures external influences such as changes in the macroeconomy that affect outcomes in both areas, and removes the effect of such changes from the estimate of the policy impact. Our comparison regions consist of the following areas: King County outside Seattle and SeaTac 4 Counties that surround King County, namely Snohomish, Kitsap, and Pierce Synthetic Seattle, defined as a set of regions in the state of Washington that have matched Seattle s labor market trends in recent years Synthetic Seattle excluding King County, to account for potential spillover of the Seattle Minimum Wage Ordinance into labor market of suburban King County Each comparison region has strengths and limitations. King County outside of Seattle and SeaTac shares a similar industrial mix with Seattle, and therefore its experience would, in ordinary times, serve as a good counterfactual for Seattle. However, using King County outside Seattle as the benchmark may not be appropriate if, as expected, the effect of the ordinance spills over to the rest of King County. As we demonstrated in an earlier report for the City of Seattle 5, the labor market for low-wage workers in Seattle is highly integrated with the labor markets of King County and surrounding counties. We found that 40% of workers in Seattle earning minimum wage live outside the city (p. 7), and among Seattle workers earning the minimum wage, only 55% work in the city (p. 8). Using Snohomish, Kitsap, and Pierce counties (SKP) attempts to address this spillover problem by using changes in the outcomes of these surrounding counties as the comparison. As one moves further away from Seattle, however, the levels and trends in employment and wages prior to the law s implementation are not as well matched to the levels and trends in Seattle. Moving from King County to SKP reduces the size of the spillover problem, but leaves us with a less similar comparison group. A limitation of our methods is that there may be other contemporaneous changes that affect Seattle s economy but not the economies of the comparison areas, or vice versa. For example, one of the city s major employers might decide to expand or contract its workforce for reasons completely unrelated to the minimum wage. Or SKP s economy might experience a significant expansion or contraction of military personnel that, again, would be completely unrelated to the minimum wage. If these differential trends continued after Seattle passed its Minimum Wage Ordinance, the differences in trends could falsely suggest an effect of the minimum wage law. 4 We exclude SeaTac from the comparison regions because it raised minimum wage to $15 for hospitality and transportation workers. 5 Klawitter, M.M., Long, M.C., and Plotnick, R.D. (2014). Who Would be Affected by an Increase in Seattle s Minimum Wage?, Report for the City of Seattle, Income Inequality Advisory Committee. 9

10 Such events cannot be controlled for in our analysis, and thus form the main threat to the internal validity of our impact estimates. To address this concern, our preferred comparison group is what we call Synthetic Seattle. This comparison group consists of an aggregate of those zip codes in the state of Washington that in the years 2005 to 2013 had similar levels and trends for the outcome and sample being studied. The details for how Synthetic Seattle is constructed are in Appendix B. We construct a second version of this measure that excludes the portions of any zip codes that lie in King County in order to address the possibility of spillover effects. Figure 33 on the next page illustrates the zip codes selected for Synthetic Seattle excluding King County and their relative sizes for one outcome (median wage rate) and one particular sample of interest (workers who had <$11 wages in quarter 0 and who were employed in quarter 6). As can be seen in this figure, the bulk of zip codes selected lie in the Puget Sound region, but some zip codes along the I-5 corridor south to Vancouver, around Spokane, and other parts of the State are likewise selected. This amalgam, while not yielding a contiguous region that is easy to contemplate as a whole, nonetheless performs very well in mimicking Seattle in the pre-minimum wage period, as is demonstrated below. No method for estimating the impact of Seattle s Minimum Wage Ordinance from secondary data is free of limitations. However, we believe our use of multiple estimation methods yield a better understanding of the likely effects of the policy and how robust the findings are to different methods. In the main text of this report, we highlight the results from Synthetic Seattle (including King County zip codes outside Seattle). We furthermore show the results for the other three comparison regions in tables in Appendix D, E, and F. If the results are sensitive to the choice of comparison region, we note this fact in the main text. Finally, note that we present our results below as point estimates, but do not include estimates of standard errors on our estimates (that is the degree of statistical uncertainty in our estimated impacts). As we move our work on this project forward, we will carefully consider the right approach for constructing these standard errors. Because the ESD data are close to the full universe of all employees, rather than a sample, one defensible position is that our results are population parameters, not sample estimates, and so standard errors are not needed. Abadie et al. (2014) note that If the researcher sees all relevant data, there is no need for inference, since any question can be answered by simply doing calculations on the data that is, for descriptive analysis of census data, standard errors are not necessary. 6 These same authors note that for causal interpretations, computation of standard errors is necessary as these standard errors capture the fact that even if we observe outcomes for all units in the population of interest, there are for each unit missing potential outcomes for the treatment levels the unit was not exposed to (abstract page). Given the lack of standard errors in this draft, some caution should be used in confidently asserting that the Minimum Wage Ordinance caused an impact of a particular size. That is, there is some degree of uncertainty in all our causal estimates. This degree of uncertainty is likely to be small for most of our estimates given the large sample sizes we are using. 6 Finite Population Causal Standard Errors, Alberto Abadie, Susan Athey, Guido W. Imbens, and Jeffrey M. Wooldridge. National Bureau of Economics Working Paper No July

11 Figure 3 11

12 An example can illustrate how our methods show both what happened? and what happened relative to what would have happened without the law? Consider an important outcome, the percentage of low-paid workers who maintain employment. To answer what happened, we first show that among Seattle workers who in the 2 nd quarter of 2014 who were employed with wages less than $11 per hour, % were employed anywhere in the state of Washington at any wage level in the 4 th quarter of This is, in the simplest sense, what happened to workers we might have expected to be most affected by the minimum wage increase. We need to answer a second question though how does this employment rate compare to prior years? We next measure business as usual by computing the same 6-quarter change in outcomes for workers in earlier years, namely the 2 nd quarter of 2005 through We find that the average rate of employment maintenance for these prior years is 62.4%. By taking the difference of these two figures (65.0% %), we find that the employment rate improved by 2.6 percentage points for Seattle s low-wage workers relative to business as usual. Seattle s low wage workers did better in this recent period than the average over the nine previous years. It is necessary to assess, however, whether this strong performance reflects the impact of the minimum wage, or other factors. It is possible that 2015 would have been a good year for Seattle s lowest-paid workers even in the absence of a minimum wage increase. To assess this possibility, and fully estimate what happened relative to what would have happened, we do the same differencing from business as usual for our comparison region, based on its usual 6- quarter changes in outcomes. Using Synthetic Seattle as the comparison region, for workers in these zip codes who in the 2 nd quarter of 2014 were employed with wages less than $11 per hour, we find that they also experienced an improvement in the employment rate relative to business as usual by 3.8 percentage points. While 2015 was a good year for the group of Seattle workers we track, it was an even better year for the workers employed in Synthetic Seattle. Finally, we take the difference of the Seattle experience from the Synthetic Seattle experience (2.6% - 3.8%) to produce our estimate of the impact of the Seattle minimum wage, which is a 1.2 percentage point decrease in the employment rate for these low-wage workers. That is, we conclude that Seattle experienced improving employment for low-wage workers, but the minimum wage law somewhat held employment back from what it would have been in the absence of the law. The Employment Security Department Data The data to answer our research questions come from administrative records kept by Washington s Employment Security Department from employers records for workers covered by Unemployment Insurance. A detailed discussion of these data, and their limitations are included in Appendix A. 7 We have intentionally chosen 2014 Q2 as a baseline in order to capture any anticipatory effects of the ordinance. This allows us to capture early adjustment both from the employers who might raise wages to their workers to comply with the ordinance, and from the workers who might switch jobs or join the labor force in anticipation of the new law. 12

13 These data include business and worker IDs, employer addresses, individual quarterly hours, individual quarterly earnings, individual quarterly industry codes, and individual quarterly unemployment benefits. Using these data elements, our study is able to analyze changes in individual employment, hours, and wage levels from quarter to quarter as well as changes in establishment level employment, payroll, and turnover. One challenge of using this dataset is that we have limited ability to properly locate the work done at large employers with multiple locations in state of Washington, such as retail or restaurant chains with company-owned stores; many of these multi-location firms file a single quarterly report to cover employees at all locations. While we can locate the address given by such multi-location firms, we are unsure whether an individual worker in these firms did his or her work in Seattle, and was thus covered by the Minimum Wage Ordinance, or in another part of the state. Consequently, we focus our analysis on single-location establishments, but separately report outcomes for the combination of single-location establishments and multi-location firms in our Appendix D, E, and F tables. For our analysis of workers, we place workers into regions (e.g., Seattle, King County Excluding Seattle and SeaTac) based on the location of the worker s primary employer, defined as the one that paid that worker the most in baseline quarter. In subsequent analyses (except where noted otherwise) we convert nominal quarterly earnings into real quarterly earnings by dividing by the Consumer Price Index for Urban Wage Earners and Clerical Workers (CPI-W). 8 All wage rates and earnings should thus be considered to be in 2 nd quarter of 2015 dollars. Outcomes Studied Wages and Workers The central outcomes that we study for wages and workers are as follows: employed, defined as having non-zero earnings from at least one job in a given quarter; earnings, defined as the sum of earnings from all jobs in a given quarter; hours worked, defined as the sum of hours worked in all jobs in a given quarter; hourly wage rate, defined as the sum of earnings from all jobs in quarter divided by hours worked in all jobs in quarter; remaining in the same region, defined as having a job in the same region as in a baseline quarter. Jobs We additionally investigate the impact of Seattle s Minimum Wage Ordinance on jobs. The Employment Security Department Data used in this report is an imperfect tool for counting the 8 We use the not seasonally adjusted CPI-W (rather than the more commonly used CPI-U) as the state of Washington indexes the state minimum wage to the CPI-W. A discussion of the differences between the two price indexes and the limitations of the CPI-W can be found in Stephen B. Reed and Kenneth J. Stewart, Why does BLS provide both the CPI-W and CPI-U?, Beyond the Numbers: Prices & Spending, vol. 3, no. 5 (U.S. Bureau of Labor Statistics, February 2014), 13

14 number of jobs in Seattle. When it comes to counting jobs, the tally of persons employed in Seattle or anywhere can and does fluctuate remarkably from day to day or month to month. Each job is counted uniquely; if a person has three jobs, they will be counted in three separate data points. The central outcomes that we study for jobs are as follows: total headcount in quarter, defined as the number of workers on payroll in the current quarter in establishments which were open in the current quarter; number of jobs at the beginning of quarter, defined as the number of workers on payroll in the current and previous quarter in establishments which were open in this and the previous quarter; this outcome most corresponds to a measure of employment published by the Quarterly Census of Employment and Wages (see Appendix A for discussion of the measures of employment in the ESD data) total hours worked in quarter, defined as the sum of hours worked in all jobs in the quarter in currently open establishments; average hours worked per job, defined as the total number of hours worked in the quarter divided by the total number of workers who were on payroll in the quarter in all currently open establishments. The data allow us to count up the total number of employer-employee relationships that existed at any point over a three-month period. We refer to this concept as headcount. Headcount may overstate the number of jobs available on a given day. Consider a small business with one wage-earning employee. If that employee quits in April and is replaced by a new employee in May, the quarterly report for the business will report a headcount of 2, even though there is never more than a single job in the business at one time. To more accurately measure the number of jobs available in Seattle at a given point in time, we focus attention on employer-employee relationships that we know to have existed at the beginning of an ESD reporting period. In the hypothetical example above, the small business in question would be counted as having one persistent job for the quarter, presuming that the employee who quit had been employed before April. Both headcount and persistent job measures are potentially useful in certain circumstances. First, we study the changes in these outcomes for all currently open establishments, which include establishments which opened since baseline quarter and contributed to job creation. Next, we study jobs, hours and total wages paid by establishments which are likely to be particularly vulnerable to the increase in minimum wage those establishments which paid more than 40% of their workers less than $11 in the baseline quarter. We track this group of establishments over time, and report the total number of jobs, hours, and wages paid for those establishments which remained open in each subsequent quarter after the baseline. Establishments We will also present results of labor market outcomes for Seattle businesses. There were nearly 21,000 single-location establishments in Seattle in 2014 Q2. The outcomes of interest are: Share of Establishments in Current Quarter Which Opened Since Baseline Quarter Share of Establishments Which Closed Since Baseline Quarter 14

15 3. Impact of Seattle s Minimum Wage Ordinance on Wages Figure 4 tallies the number of workers primarily employed at single-location Seattle businesses with wages in specific one-dollar increments in the baseline quarter during which the Seattle minimum wage was passed (2 nd quarter of 2014) and six quarters later (4 th quarter of 2015). This figure includes only workers earning less than $25 per hour as we do not anticipate Seattle s Minimum Wage Ordinance to have cascading effects on wages above this level. Tallies from the earlier period are represented by solid bars and from the more recent period by transparent outlines. Note that in the baseline quarter, the state of Washington s minimum wage was $9.32 per hour, and we see that a large share of workers in this figure were earning between $9 and $10 per hour. 9 Six quarters later, we see a clear decrease in the share of Seattle s workers earning between $9 and $11 per hour, which would suggest that the minimum wage caused an upgrading of worker s wages. We see sizable gains in the share of workers earning between $11 and $16 per hour. 10 The figure clearly shows a stark reduction in the number of workers earning between $9 and $10 per hour; there were 19,056 such workers in mid-2014 but only 7,330 at the end of The number of workers earning between $10 and $11 per hour fell from 21,470 to 15,469. As the number of workers earning under $11 shrank, the number earning $11-13/hour rose significantly, and the number earning $13-19/hour rose enough to be noticeable on this chart. The number of workers with average hourly wages in each increment above $19/hour shows very little change. A direct comparison of the size of the bars in Figure 4 shows that the total number of workers represented declined over time. In other words, the reduction in the number of workers earning under $11/hour exceeds the increase in workers earning $11-19/hour shown here. Over a time period where the Seattle economy added tens of thousands of jobs, the number of workers employed at single-location firms earning wages under $25 declined by 7,385, or about 3.3% of the baseline number, with all of that change accounted for by a decline in workers earning under $19/hour. 9 There is a small share of workers who appear to be earning less than $9 per hour. We caution readers to be aware that this result could be due to inaccuracies in the employer s report of the employee s number of hours worked in the quarter. Inaccurate reports that overstate hours would result in our computed hourly wage rate being too low. This could also reflect accurate reporting of hourly wage rates for certain employees who are exempt from the minimum wage law. 10 Note that in our report in April, 2016 based on surveys of firms, we noted Roughly one-fifth of all firms reported a wage level higher or lower than the Ordinance required, with almost one in ten firms (11.9%) indicating the Ordinance required them to pay a $15 per hour minimum wage as of April 1, (The Seattle Minimum Wage Study Team Report on Baseline Employer Survey and Worker Interviews. Seattle. University of Washington.) 15

16 Figure 4 Figure 4: Nominal Hourly Wage Rate Distribution of Seattle s Workers (for those earning <$25 per hour) Figure 5 charts the proportion of workers with (inflation-adjusted) wages under $11 per hour from 2005 through the end of The period between late 2007 and early 2009, identified as the beginning and end dates of the most recent recession, is marked in gray on this graph. This figure is not seasonally adjusted; the regular dips and peaks reflect seasonal fluctuations in the economy. In this figure we show that the share of Seattle s workers earning less than $11 per hour was relatively stable prior to the passage of the Minimum Wage Ordinance, aside from seasonal fluctuations. 12 It began falling rapidly after passage of the law. This result suggests that firms began responding immediately to the new ordinance by raising wages in anticipation of the ordinance coming into force. 12 Low-wage employment peaks are generally in summer months during tourist season and summer closure. The shaded region on Figure 5 shows the period of the Great Recession according to the National Bureau of Economic Research ( 16

17 Figure 5 Figure 5: Temporal Changes in the Wage Rate Distribution in Seattle and Snohomish, Pierce, Kitsap Counties Between 2005 and However, when we do the same interrupted time series analysis for workers in Snohomish, Pierce, and Kitsap counties, we see a very similarly timed drop in the share of workers earning less than $11 per hour. The similarity of the timing of the decline suggests that broader macroeconomic forces, and to some extent typical seasonal fluctuation, may have been the cause of the decline in the share of workers earning less than $11 per hour. The next sections of this report will endeavor to more carefully and systematically rule out these other possible explanations. Finally, it is worth reflecting on the differences between the time series evidence for Seattle versus Snohomish, Pierce, and Kitsap counties. Historically, Seattle has had a lower share of workers earning less than $11 per hour than in SKP, and the variability and seasonality of the share of workers earning less than $11 has been less in Seattle than in SKP. These results suggest that SKP may not be a perfect comparison group for Seattle more specifically, the macroeconomy may be affecting SKP differently than Seattle. This concern motivates our use of Synthetic Seattle as our favored comparison region, as Synthetic Seattle is chosen so as to specifically match Seattle s pre-minimum wage levels and trends in outcomes. 17

18 Key findings: The distribution of wages shifted as expected. The share of workers earning less than $11 per hour declined sharply. Decline began shortly after the ordinance was passed. However, similar declines were seen outside of Seattle, suggesting an improving economy may be the cause of the change in the distribution of wages. 18

19 Median Wage 4. Impact of Seattle s Minimum Wage Ordinance on Low-Wage Workers While the prior section tells us that the distribution of wages changed in the expected manner (i.e., a large reduction in the share of workers earning less than $11 per hour). This section, builds on the prior section by evaluating how this change in the wage distribution affected lowwage workers. In this section, we follow cohorts of low-wage workers longitudinally to see how their labor market outcomes changed. We focus on five key outcomes: Wages, Employment, Hours Worked, Quarterly Earnings, and Remaining Employed in Seattle. Wages: Figure 6 shows the median wage for a post-policy cohort of workers who in the baseline quarter (2 nd quarter of 2014) were working in Seattle and earning less than $11 per hour. We follow this cohort of workers forward in time for the next six quarters and compute their median wage (with $0 per hour imputed for those who are not working in the state of Washington in that quarter). We find that median wage for this cohort of workers increased from $9.96 to $11.14 over the next six quarters. While it would be intuitive to attribute this increase to the Minimum Wage Ordinance, our methodology requires us to compare what happened to estimates of what would have happened in the absence of the Minimum Wage Ordinance. Figure 6 Figure 6: Median Wages for Cohort of Seattle Workers Who Earned < $11 Per Hour in 2 nd Quarter of 2014 $11.50 $11.00 Minimum Wage Ordinance Passed Minimum Wage Ordinance Started $10.50 $10.00 Seattle, 2014 Cohort (Post-Policy) $9.50 $ Quarters After Baseline (2 nd Quarter 2014) 19

20 Median Wage Figure 7 repeats this analysis for the prior nine cohorts (2005 to 2013), with the baseline always beginning in the 2 nd quarter of the year. We characterize business as usual as the average of the 2005 to 2013 cohorts. There are two things to note from this chart. First, the increase in median wages experienced by the 2014 cohort between the 3 rd and 4 th quarter post baseline (i.e., at the time that the minimum wage started) was far outside of the historical norm. This result provides strong evidence that the minimum wage likely caused these workers wages to rise. Second, while there was some improvement in recent years, as demonstrated by 2011, 2012, and 2013 cohorts, the experience of the 2014 cohort vastly exceeds these recent cohorts beginning at the time the minimum wage law went into effect. Figure 7 Figure 7: Median Wages for Cohorts of Seattle Workers Who Earned < $11 Per Hour in 2 nd Quarter of 2005, 2006,, and 2014 $11.50 Minimum Wage Ordinance Passed Minimum Wage Ordinance Started $11.00 $10.50 $10.00 $ Cohort 2006 Cohort 2007 Cohort 2008 Cohort 2009 Cohort 2010 Cohort 2011 Cohort 2012 Cohort 2013 Cohort Pre-Policy Average 2014 Cohort $ Quarters After Baseline (2 nd Quarter ) Next, to assess whether the result shown above is due to the minimum wage law or macroeconomic forces, we repeat this analysis for Synthetic Seattle and report the differences with Seattle. Figure 8 presents the results. First, note that Synthetic Seattle is a very good match for Seattle in the pre-policy year, as can been seen by the similarity of the solid and dashed purple lines. Second, note that the strong economy can be seen in Synthetic Seattle by comparing the dashed black line for post-policy Synthetic Seattle with the dashed purple line for the pre-policy average cohort in Synthetic Seattle. Thus, some of the gain in wages seen in Seattle for these low-wage workers is likely due to the strong economy. However, Seattle begins to diverge and exceed Synthetic Seattle just as Seattle s Minimum Wage Ordinance comes into effect. 20

21 Median Wage Figure 8 Figure 8: Median Wages for Seattle and Synthetic Seattle Workers Who Earned < $11 Per Hour in the 2 nd Quarter $11.50 Minimum Wage Ordinance Passed Minimum Wage Ordinance Started $11.00 $10.50 $10.00 Seattle, 2014 Cohort (Post-Policy) Synthetic Seattle, 2014 Cohort (Post-Policy) Seattle, Pre-Policy Average $9.50 Synthetic Seattle, Pre-Policy Average $ Quarters After Baseline (2 nd Quarter 2014 for Post-Policy) We find that these low-wage workers median hourly wage increased by $1.18. Business as usual for such workers (as given by the historical change in Seattle for such workers) is a $0.14 loss in hourly wage rate. Thus, relative to Seattle s own history, such workers median hourly wage rate rose by $1.32. A sizable portion of this gain, however, can be attributed to the strong macroeconomy. Therefore, we estimate the effect of Seattle s Minimum Wage Ordinance as a $0.73 increase in these workers median hourly wage rate (based on the difference-in-differences between Seattle and Synthetic Seattle). Figure 9 graphically shows our estimate of the impact of the Minimum Wage Ordinance (i.e., the difference-in-differences) on median wages for these low-wage workers. In the remainder of this report, in the interest of brevity, we evaluate changes between the baseline (t=0) and six quarters hence (t=6), but do not report the intervening quarters (t=1 to t=5). In the main text of the report, we discuss the central findings. The full results are in Appendix D. 21

22 Estimated Impact on Median Wage Figure 9 Figure 9: Estimated Impact of Seattle s Minimum Wage Ordinance on the Median Wages for Seattle Workers Who Earned < $11 Per Hour in the 2nd Quarter of 2014 $1.50 $1.25 Minimum Wage Ordinance Passed Minimum Wage Ordinance Started $1.00 $0.75 $0.50 $0.25 Difference-in- Differences Between Seattle and Synthetic Seattle $ ($0.25) ($0.50) Quarters After Baseline (2 nd Quarter 2014 for Post-Policy) Our evaluation for other labor market outcomes for these low-wage workers parallels the above analysis of wages. Employment: While the intended effect of the Minimum Wage Ordinance (i.e., raising low-wage workers wages) appears to have been successful, there appears to have been some negative impacts on these worker s rates of employment and hours worked. As noted previously, the rate of employment of these workers increased by 2.6 percentage points. However, the comparison regions all experienced even better employment rate increases (3.8% for Synthetic Seattle, 3.9% for Synthetic Seattle Excluding King County, 3.5% for SKP and 2.9% for King County Excluding Seattle and SeaTac). Thus, it appears that the Minimum Wage Ordinance modestly held back Seattle s employment of low-wage workers relative to the level we could have expected. Hours worked: Hours worked shows a similar pattern. Among workers earning less than $11 per hour at baseline in Seattle, hours worked increased by 12.2 relative to business as usual. 13 So, again, Seattle s employment situation for low-wage workers improved after the Minimum Wage Ordinance was passed. Hours worked increased, however, by more in the comparison regions (16.4 for Synthetic Seattle, 13.0 for Synthetic Seattle Excluding King County, 21.5 for SKP and 22.5 for King County Excluding Seattle and SeaTac). Thus, on balance, it appears that the Minimum Wage Ordinance modestly lowered hours worked (e.g., 4.1 hours per quarter relative to Synthetic Seattle, or 19 minutes per week). 13 On average, these low-wage workers worked 276 hours in the 2 nd Quarter of

23 Quarterly earnings: Given the gains in hourly wage rates, but the offsetting declines in rates of employment and hours for low-wage workers relative to similar workers in comparison regions, it should not come as a surprise that we find small effects on low-wage workers quarterly earnings. Quarterly earnings for Seattle workers earning less than $11 per hour at baseline increased by $463 relative to business as usual again, this was a good six quarter stretch for low-wage Seattle workers. Most if not all of this gain in quarterly earnings can be explained by the robust economy rather than the Minimum Wage Ordinance. For example, similarly situated workers in Synthetic Seattle saw their quarterly earnings rise by $391 which would suggest that the Seattle Minimum Wage Ordinance increased quarterly earnings by $72 (or $5.54 per week). This modest positive result is sensitive to the comparison region; King County outside Seattle and SeaTac and SKP saw quarterly earnings rise by $531 and $495, respectively, for similarly situated workers, which would suggest that the Seattle Minimum Wage Ordinance decreased quarterly earnings by $68 or $32 respectively (or $5.22 or $2.43 per week, respectively). The bottom-line here is that the effect of the Seattle Minimum Wage Ordinance on low-wage workers earnings is ambiguous and likely fairly small. Effects for those who remained employed: When we restrict our analysis to workers who earned less than $11 per hour in the baseline quarter and were employed anywhere in Washington six quarters hence we find clearly positive impacts of Seattle s Minimum Wage Ordinance. For such workers in Seattle, median hourly wage rates increased $1.03 relative to business as usual and this growth in wages exceeded that experienced by all comparison regions; using Synthetic Seattle as the comparison region, for workers who kept their jobs, we find that the impact of the ordinance was to raise these workers median hourly wages by $0.31. These Seattle workers increased their hours worked by 7 hours per quarter relative to baseline, but this was less than all comparison regions. Using Synthetic Seattle as our comparison region, we find that the impact on hours for those who kept their jobs was -3.2 hours per quarter, or 15 fewer minutes per week. From the perspective of the workers, a modest cut in hours might be considered a good thing if it doesn t adversely affect earnings. Yet, we find that earnings for these Seattle workers who kept their jobs increased by $542 per quarter (a sizable gain given their baseline quarterly earnings of $2,524) and this increase was greater than all comparison regions, which suggests that Seattle s Minimum Wage Ordinance raised these workers earnings by $164 using Synthetic Seattle as our comparison region, or $13 per week, a 6.5% increase relative to earnings in the baseline quarter. Thus, if you kept working, you were modestly better off as a result of the Minimum Wage Ordinance, having $13 more per week in earnings and working 15 minutes less per week. Remaining in Seattle: While low-wage workers may be finding employment opportunities, it is not necessarily the case that they are finding such work in Seattle, per se. Among those workers who earned less than $11 per hour in the baseline quarter and who were employed somewhere in Washington six quarters hence, only 70.1% of them are still working in Seattle. While this figure may sound like a large drop, the typical rate found for the prior nine cohorts was 73.4% of such workers remaining in Seattle. Thus, relative to business as usual, we find a decline of 3.3 percentage points in the share of such workers remaining in Seattle. We compare this relative to decline to that experienced in Snohomish, Kitsap, and Pierce counties (as remaining in Synthetic Seattle is a not well defined concept). We find that, relative to business as usual, there was a decline of 0.5 percentage points in the share of such low-wage workers remaining employed in SKP rather than elsewhere in the state. Differencing these results, we conclude that the Seattle 23